Sports Betting Odds Dataset
Two CSV files, 9,805 rows of real odds captured from 23 distinct sources: sportsbooks, betting exchanges, and prediction markets including Polymarket and Kalshi. Free under CC BY 4.0, no account needed with us.
What is in it
worldcup_2026_odds_snapshot.csv6,132 rows · 20 sources
The full odds board for the 2026 FIFA World Cup, captured on 13 July 2026 during semifinals week (Argentina vs England, France vs Spain). 99 market types, from moneylines and Asian handicaps through to correct score and outrights.
mlb_odds_snapshot.csv3,673 rows · 14 sources
A same-day MLB slate across 14 of the same sources: moneylines, run lines, totals, and derivatives.
Schema
One row is one price, on one selection, at one source, at capture time. Both files share the same 17 columns.
idsportsbookevent_idsportleaguehome_teamaway_teammarket_typeselectionselection_typeodds_americanodds_decimalodds_probabilitylineevent_start_timeis_livetimestampKnown quirks
Worth reading before you join on anything. These are real properties of the data, not defects we are hiding.
idis a price record id and is not unique across all rows. Sixteen ids are each shared by two World Cup rows with distinct selections. Identify a single price byevent_id+sportsbook+selection.- Prediction-market sources (Kalshi, Polymarket) encode markets as yes/no questions, so their
home_teammay hold the full question text. odds_probabilityis implied probability with the vig still in it. Remove the overround yourself if you need fair odds.- These are point-in-time snapshots, not a live feed. For current prices use the API.
Licence
Released under CC BY 4.0. Use it commercially, redistribute it, build on it. The only condition is attribution to SharpAPI.